We are using 3x DL380 G7 with Windows 2012 R2 Hyper-V Cluster. In order to get more performance out of it we added 2 HPE NC523sfp 2P 10GB NICS per host. The ISCSI traffic worked as a charm.
So we had a consultant in to try and use the 2 remaining free 10GB port to improve LAN and Cluster network speed.
The HP Switch stack And ports are configured to use LACP. Host1 paused, he removed the 1GB NICS from the LAN Team and replaced them with the 10GB NICS. He tried to make the virtual switch converged so it would be used for VM traffic, management
and cluster traffic. Everything seemed to go smooth, no errors and no restart needed. After resuming VM's were live migrated back without problems.
So he started on host2 same procedure. After resuming and life migrating VM's, Host1 suddenly BSOD with stopcode 133 PDC watchdog violation error. The driver is the latest HPE november 2015 driver for this QLogic NIC.
He removed the VLAN tag and virtual switch configuration so it would only be used for VM LAN traffic. Management and Cluster, Live Migration back to their own 1GB NICs. This time it remained stable so also Host3 was done. Live Migrating, pausing and resuming and the nightly backup of the VM's went without problems.
Yesterday I installed the July update and after completing successfully updating all three hosts, host3 started to BSOD under normal VM workload. I paused it to investigate further and this night also Host1 and Host2 went BSOD under the VEEAM VM backup.
I have read that there are VMQ issues with these older NICS, NIC teaming due to overlapping processors. According to the consultant that only relates to switch independent teaming. We use LACP and load balancing mode is set to hashtag instead of dynamic.
For now I disabled VMQ on the NIC team and team NICS.
Are there know VMQ issues with these older NICS, W2012R2 Hyper-V Host and W2008R2 VM's also when using LACP?
TIA,
Fred
